Head nodding and head shaking gestures in children’s early communication

Abstract

Children’s spontaneous head nodding and head shaking gestures are examined in the context of mother-child play. Observations at 14, 20 and 32 months show that frequency of head gestures increases with age, with a pronounced increase in nodding between 20 and 32 months. Over time, children increasingly use head gestures in combination with speech, though isolated head nods continue to predominate at 32 months. Pragmatic analysis of children’s head gestures reveals that they serve a small set of communicative functions during early childhood. Children’s use of a head gesture for communication at 14 months was a significant predictor of pragmatic flexibility at 32 months, even when controlling for 14-month language production and maternal language input. The significance of head gestures in children’s emerging communication is discussed.

Keywords

conventional gestures intentional communication parent-child interaction pragmatics preverbal communication

Gestures are among the earliest emerging forms of intentional communication for interacting and sharing meaning with others. With a pointing gesture, for example, preverbal infants can intentionally direct another person’s attention to an object (proto-declaratives), use someone as a tool in achieving a goal (proto-imperatives; Bates, Camaioni, & Volterra, 1975), and provide someone with information (Liszkowski, Carpenter, Striano, & Tomasello, 2006; Tomasello, Carpenter, & Liszkowski, 2007). In many cultures, the head nod and head shake (subsequently referred to as head gestures) are ubiquitous conventional gestures, or emblems, supplementing or replacing the words ‘yes’ and ‘no’ in response to questions, and signaling attitudes like agreement and disagreement, and approval and disapproval. Based on parent reports from a large U.S. sample, the ages at which most children produce head shakes (12 mos.) and head nods (14 mos.) precede those for the corresponding verbal forms ‘no’ (15 mos.) and ‘yes’ (19 mos.; Fenson et al., 1994). Given their relatively early emergence and their many potential communicative functions, head gestures may be particularly powerful tools for communication during early childhood.

Conventional head gestures have often been overlooked in developmental studies of gesture production, with some exceptions (e.g., Crais, Douglas, & Campbell, 2004; Guidetti, 2005). A comprehensive study of language development by Fenson et al. (1994) considered only the timing of emergence, and not the frequency or function of children’s head gestures. Gesture researchers have tended to group head gestures with other conventional gestures, like waving, in their analyses (e.g., Capirci, Iverson, Pizzuto, & Volterra, 1996; Iverson, Capirci, Longobardi, & Caselli, 1999; Iverson & Goldin-Meadow, 2005), or to focus on other categories of gestures, such as iconic or symbolic gestures (e.g., Acredolo & Goodwyn, 1988) and pointing (Carpenter, Nagell, Tomasello, Butterworth, & Moore, 1998). Specific consideration of conventional head gestures may be critical for estimating the child’s early communicative competence. In a study of one-year-olds’ expression of communicative intent, Yont, Snow, and Vernon-Feagans (2001) observed that infants’ head nods in parent-infant interactions can raise an unintelligible vocalization to the status of a speech act by clarifying intent. Thus, closer examination of how often and for what purposes children use head gestures warrants further research.

In a study specifically focused on head gestures and their verbal counterparts, Guidetti (2005) found that French children’s agreement and refusal messages produced while interacting with their mothers were largely conveyed in gesture alone at 16 months (72% of messages). Guidetti further reports that among two- and three-year-old French children, verbal-only agreement and refusal messages outnumbered both isolated head gestures and gesture-speech combinations, though children in each age group used all three message forms. This pattern matches previous findings of a shift from the predominance of isolated gestures, with or without non-word vocalizations, to that of gesture-speech combinations and isolated words from 12 to 27 months of age (Carpenter, Mastergeorge, & Coggins, 1983; Wetherby, Cain, Yonclas, & Walker, 1988). Based on observational studies of parent-child communication, mothers’ use of isolated gestures, including conventional gestures, is relatively rare (Iverson et al., 1999; Rowe, Özçalışkan, & Goldin-Meadow, 2008). Thus, conventional head gestures in isolation may be particularly evident in the communication of young toddlers, whereas older children and adults more often merge or replace them with verbal language. However, this developmental progression has yet to be tested for head gestures in an English-speaking sample.

Bates et al. (1975; Bates, Benigni, Bretherton, Camaioni, & Volterra, 1979) argued that communication through gesture and through verbal language rely on an overlapping set of cognitive systems. Verbal language skills are correlated with at least four early gestures (the Gesture Complex: giving, showing, communicative pointing, and ritualized requests) perhaps because they share an underlying structure related to the use of conventional signals to communicate. This perspective is supported by evidence that variation in early use of gesture and gesture-speech combinations predicts later verbal language outcomes (Capirci et al., 1996; Laakso, Poikkeus, Katajamäki, & Lyytinen, 1999; Morford & Goldin-Meadow, 1992; Rowe et al., 2008; Sauer, Levine, & Goldin-Meadow, 2010). For example, Rowe and colleagues (2008) report that the breadth of children’s gesture ‘vocabulary’ at 14 months, that is, the number of distinct meanings conveyed through gesture, predicts children’s verbal vocabulary size at 42 months, above and beyond the effects of 14-month word-type production. Further, the age at which children produce speech+gesture combinations that convey two meanings (e.g., point to cup+‘mommy’) is predictive of the onset age for producing two-word utterances (Capirci et al., 1996; Iverson, Capirci, Volterra, & Goldin-Meadow, 2008; Iverson & Goldin-Meadow, 2005). Rowe and Goldin-Meadow (2009) report that aspects of gesture use at 18 months selectively predict related language skills at 42 months. First, gesture vocabulary at 18 months (the number of different meanings the child conveyed via gesture) predicts later vocabulary size, but not syntactic skill. In contrast, the number of sentence-like speech+gesture combinations produced at 18 months predicted 42-month syntax skill, but not vocabulary size. Thus, analysis of early gesture use offers a window into various features of children’s language development, and into the continuities between early and later modes of expressing intent.

The current study uses and extends this continuity perspective of gestural and verbal communication development to examine the relationship between early use of head nodding and head shaking gestures and later pragmatic verbal skills. We hypothesize that the production of conventional head gestures to express a variety of communicative intents, such as to agree and disagree, to respond in the affirmative and negative to yes/no questions, and to agree and refuse to carry out requested acts, might serve as a foundation from which to build these pragmatic skills in verbal language. A prediction that follows from this view of language development is that children who use head gestures in the service of these communicative goals are well on their way to using verbal language to convey a similar array of communicative intents. Indeed, Bates and colleagues (1979) found that the production of ritualized refusals, that is, shaking the head ‘no’ to make a refusal, was positively but not significantly related to later language skills. They suggested that the child who protests with conventional signals (i.e., shaking his head ‘no’) is probably more sophisticated in all forms of communication than a child who refuses using non-conventionalized signals, such as turning away and fussing. The non-significance of this finding may have been due to the limited age range they examined, 9 to 13 months, as head shakes typically emerge at the end of this period (Fenson et al., 1994). The current study aims to test the positive relationship between early head gesture use and later pragmatic language skills, based on a longitudinal analysis of children’s communication development from 14 to 32 months.

The current study

The current study examines young children’s spontaneous production of conventional head gestures while interacting with their mothers. The research questions are:

With what frequency do typically developing children in the U.S. produce head gestures, with and without verbal utterances, while interacting with their mothers?

What communicative functions do young children’s head gestures serve?

Is early head gesture production related to verbal language development, particularly in the domain of pragmatics?

Method

Participants

Participants were a subset of parent-child dyads included in the New England corpus of the Child Language Data Exchange System (Pan, Imbens-Bailey, Winner, & Snow, 1996; Snow, Pan, Imbens-Bailey, & Herman, 1996). Data were originally collected from 52 dyads longitudinally, in the 1980s, when children were approximately 14, 20, and 32 months. Children with indications of medical or other developmental problems were excluded. A subset of 33 dyads (n = 16 female; n = 17 male) with all three waves of video and transcript data was used for the current study. Participating caregivers were all mothers, with the exception of one father at one time point (Time 3, female child). The original sample was drawn from English-speaking, predominantly White families living in New England, of lower-middle and upper-middle class households (MacWhinney, 2000). The mean Hollingshead’s Four Factor Index of Social Status score (Hollingshead, 1975) for the current sample, measured at Time 1, ranged from 32 to 66 (M = 55.67, SD = 10.14).

Procedure

Parent-child dyads were videotaped during a semi-structured play session in a laboratory setting. At 14 and 20 months, the sessions began with a 5-minute warm-up period, during which several toys were available for free-play (e.g., small football, squeaky duck, book, jack-in-the-box, toy car). Dyads then played with four standard toys: a ball, a cloth for peek-a-boo, crayons and paper, and a book. At 32 months, the warm-up period was omitted, and the standard toys included a book, hand puppets, crayons and paper, and a toy house. At each time point, parents were instructed to use the standard toys, each contained within a separate box, in succession for a total of about ten minutes. Within the ten minutes, participants were free to play with each item for as long as they chose. Sessions were terminated after the parent had tried to engage the child in all four activities, yielding sessions that ranged from 9 to 25 minutes. The average session length was 15:27 at Time 1, 13:58 at Time 2, and 14:36 at Time 3.

Sessions were video-recorded on VHS tapes. Given the original study’s primary interest in child behavior and language, the camera tracked the child when it was not possible to keep the activity of both participants in range. For the following analyses, calculation of head gesture frequency rates (i.e., gestures per minute) was based on on-screen time for children, which averaged 15:25 at Time 1, 13:46 at Time 2, and 13:49 at Time 3.

Transcription and coding

Parent-child communication had previously been transcribed using the CHAT conventions of the Child Language Data Exchange System (MacWhinney, 2000; MacWhinney & Snow, 1985, 1990). Transcripts are publicly available through the CHILDES system, and videos are archived at the Harvard Graduate School of Education. Parent and child communication was segmented into verbal utterances, based on intonation contour and pause duration. Several measures of child and maternal language in the overall sample were generated by the original researchers (Pan et al., 1996; Snow et al., 1996). Child vocabulary skill was indexed by the number of word types (i.e., unique words) and word tokens (i.e., total words) that children spontaneously produced, based on the first 10 minutes of play with the materials. At Time 1 and 2, the 10-minute excerpt included the first five minutes of the warm-up period and of the play with standard toys. Descriptive statistics for these verbal language data are shown in Table 1. Time 1 measures of maternal verbal language included the total number of word tokens produced in the same 10-minute excerpt (M = 562.97, SD = 182.94), and the type/token ratio for the session, which indexes the uniqueness of mothers’ vocabulary (M = .35, SD = .04).

Table 1.

Descriptive statistics for child language measures (N = 33).

	Session
	1		2		3
	Mean	SD	Mean	SD	Mean	SD
Age in months	14.19	0.25	20.02	0.54	30.71	1.62
Word types in 10 min.	3.21	2.90	28.27	15.06	86.70	26.45
Word tokens in 10 min.	6.45	6.94	63.48	39.36	217.48	78.41

In the original transcripts, non-word vocalizations and freestanding conventional gestures were also captured. Gestures that accompanied verbal or vocal acts were systematically transcribed only when they clarified the communicative attempt (Ninio, Wheeler, Snow, Pan, & Rollins, 1991). Given these guidelines, several instances of head gestures were not recorded in the transcripts. Thus, for the current study, the videos were recoded by the first author, and all mother and child head gestures and the verbal utterances they accompanied were documented.

To assess inter-coder reliability for child head gesture coding, 10% of all sessions were double coded by a research assistant. Reliability expressed as a simple percent agreement between the primary coder and the second coder was 84%. This percentage represents the number of child head gestures identified in common by both coders, divided by the total number of head gestures identified across the two coders. An alternative calculation of percent agreement has been used in previous observational studies of gesture production (e.g., Capirci et al., 1996; Guidetti, 2005), and is drawn from Sears, Rau, and Alpert (1965) and Kratochwill and Wetzel (1977). To generate this value, the number of gesture instances identified in common by the two coders is doubled, and divided by the sum of the number of instances captured by each coder. Using this method, the percent agreement for identifying head gestures was 91%. Disagreements were resolved by discussion between the coders.

The investigators of the original child language study coded the intent of children’s communicative behavior using the Inventory of Communicative Acts-Abridged (INCA-A; Ninio et al., 1991; Ninio & Snow, 1996; Ninio, Snow, Pan, & Rollins, 1994). The INCA-A is a comprehensive taxonomy of communicative acts that occur in parent-child interactions. It is derived from a social interactionist perspective of development, which emphasizes the embedding of children’s learning within the context of social and communicative interactions. This coding scheme captures, for example, whether children’s communicative acts serve as responses to yes/no questions, or as agreement and disagreement messages. Unlike other schemes used to classify early emerging intentions (e.g., Bruner, 1981; Carpenter et al., 1983; Guidetti, 2005; Wetherby et al., 1988), this taxonomy accounts for a wide range of communicative intents that reflect social and cognitive achievements during and beyond the first year of life. The inventory categorizes communication at two levels: the social interchange, which considers the function of multiple rounds of talk, and the individual speech act. A measure of pragmatic flexibility derived from this coding scheme counts the total number of interchange-speech act combinations expressed two or more times in children’s verbal language. Pragmatic flexibility at Time 1 was used as a measure of early verbal language skill (M = 1.94, SD = 2.67). Pragmatic flexibility at Time 3 was used as the key measure of child language outcomes (M = 20.64, SD = 5.71).

For this study, the communicative intents expressed through children’s head gestures were coded by the first author using the speech act level of the INCA taxonomy. A second coder applied this coding scheme to a random subset of 15% of head gesture instances, and identified whether each gesture occurred in isolation or with speech. Inter-rater reliability of the communicative act coding, as assessed by Cohen’s kappa, was 0.83. Simple percent agreement for identifying whether gestures occurred with or without speech was 93%. Disagreements were resolved by discussion between the two coders.

Results

Frequency of head gestures

To address the first research question, we examined the number of children producing each head gesture type, and the frequency of head gestures observed in each play session. To account for variation in session lengths, frequency rates were calculated as the number of head gestures in a given session divided by the number of minutes of on-screen observation time in that session. Frequency rates were multiplied by 10 for ease of interpretation (i.e., number of gestures per 10 minutes). Frequency rates of zero were included for children who did not produce a head gesture during a given session. In total, four children did not produce a head gesture in this play context at any of the three time points.

Both the number of children who produced each head gesture, and the mean frequency rates for gesture production, increased over time (Table 2). Because frequency distributions were highly skewed, non-parametric tests (Wilcoxon Signed-Rank Tests) were used to compare rates between time points. This analysis revealed that for both gesture types, the difference in frequency rates between 14 and 32 months was significant (Nods: z = −4.37, p < .001; Shakes: z = −3.43; p = .001). For head nods, the difference between 20-month and 32-month frequency rates was also significant (z = −3.70, p < .001). For head shakes, the difference between 14- and 20-month frequency rates approached statistical significance (z = −1.79, p = .074). Thus, while children used both head gestures with increasing frequency between the first and third observation, the change in production rates was more pronounced between the earlier time points for head shakes, but between the later time points for head nods.

Table 2.

Frequency of head gesture production by gesture type and child age (N = 33).

	Frequency per 10 min.
	Age	N children	n gestures	M	SD	Min	Max
Head Shakes	14	10	14	.27	.43	0	1.43
	20	14	40	.86	1.72	0	8.70
	32	20	56	1.23	1.40	0	4.90
Head Nods	14	5	11	.21	.58	0	2.48
	20	9	30	.67	2.00	0	11.09
	32	25	116	2.57	4.32	0	23.29

Wilcoxon signed-rank tests were also used to compare the relative frequency of head nods and head shakes at each time point. At the first two time points, children produced both head gestures at similar rates. At 32 months, the difference between head nod and head shake frequency rates approached significance (z = 1.66, p = .098), with head nods occurring somewhat more frequently than head shakes.

The majority of head nods and head shakes at 14 and 20 months were produced without speech (Figure 1). At 32 months, most head shakes were accompanied by words, while head nods continued to be produced primarily in isolation. Thus it appears that, overall, children tend to use head gestures more often without speech in this period of development. By 32 months, head shakes, more so than head nods, were coupled with verbal utterances.

Figure 1.

Proportion of head gestures observed with and without accompanying speech, by gesture type and child age.

Functions of head gestures

To address the second research question, the communicative intents expressed through children’s head nodding and head shaking gestures were analyzed, and results are presented in Table 3. Overall, children’s head shakes primarily served three communicative functions. These were (a) refusing to do an act suggested or proposed by the mother (e.g., in response to mother holding out peek-a-boo cloth and suggesting, ‘You do it’); (b) answering ‘no’ in response to mothers’ yes or no questions (e.g., using a Cookie Monster puppet, mother asks ‘Do you have any cookies little boy?’ Child shakes head ‘no’); and (c) reinforcing statements that included syntactic negations (e.g., playing with a toy house, child shakes his head while stating, ‘That room doesn’t have, um, any toys’). Statements with negations also included those that occurred in response to questions other than those in a yes/no format (e.g., in response to ‘Who’s this?’, child shakes his head while saying ‘I don’t know’).

Table 3.

Number of children who produced head gestures representing each communicative act, and the proportional occurrence of communicative acts, by gesture type and child age.

	14 months		20 months		32 months
	n children	% of gestures	n children	% of gestures	n children	% of gestures
Head Shakes
Refuse to do	6	50.0	9	65.0	8	19.6
Answer ‘no’ to a yes/no question	2	14.3	4	17.5	10	35.7
Make a statement with a negation (e.g., ‘no’, ‘not’, ‘don’t’)	1	7.1	2	5.0	10	30.4
Answer ‘yes’ to a yes/no question (double negative)	0	0.0	2	5.0	5	12.5
Prohibit from doing	0	0.0	1	2.5	0	0.0
Other	3	28.6	2	5.0	1	1.8
Head Nods
Agree to do an act	3	36.4	5	33.3	9	12.1
Answer ‘yes’ to a yes/no question	2	27.3	4	33.3	19	48.3
Agree with mother’s proposition	2	18.2	1	3.3	6	20.7
Answer call/Acknowledge	0	0.0	1	13.1	2	2.6
Make an affirmative statement	0	0.0	2	13.3	10	14.7
Other	2	18.2	1	3.3	1	1.8

Head nods were primarily used in the context of five communicative acts. These were (a) agreeing to do an act suggested or proposed by the mother (e.g., in response to mother’s request ‘Do you think you could draw something for Kate?’); (b) answering ‘yes’ in response to mothers’ yes or no questions (e.g., mother reads ‘I brush my teeth when I get up’ then asks ‘Do you do that?’, child nods); (c) answering mothers’ calls for attention (e.g., child nods and says ‘mhmm’ to acknowledge or show attentiveness to mother’s utterances); (d) agreeing with propositions made by mothers (e.g., mother corrects child’s identification of a picture by saying ‘that’s a round soap’, to which child responds with ‘yeah’ and a head nod); and (e) reinforcing affirmative statements, including single nods emphasizing a certain word (e.g., looking at a picture book, mother prompts ‘and a …’, the child responds ‘knife’ while nodding his head once.). Affirmative statements also included responses to questions other than those in a yes/no format (e.g., mother asks ‘sneakers or sandals?’, child nods while saying ‘sandals’).

Table 2 also presents the number of children who used a head gesture for each function at least once, by gesture type and age. As a group, the children produced head gestures for a similar range of functions at each time point. The three communicative act types not observed until the second time point included answering in the affirmative using a head shake (e.g., confirming a negative response), and using head nods while making affirmative statements and answering calls for attention by the mother. Rather than a sharp increase in the number of different communicative acts performed via head gesture, this analysis generally revealed increases in the number of children producing each communicative act type through head gesture over time.

Table 2 also displays the percentage of observed head gestures that represented each type of communicative act, by age and gesture type. This descriptive analysis revealed that at 14 and 20 months, head shakes predominantly conveyed refusal (50% and 65% of head shakes, respectively). By 32 months, refusals were no longer predominant; although head shakes were used as refusals by a similar number of children at 20 and 32 months, their relative frequency in children’s head gesture production declined to 19.6% at Time 3. Head nods were primarily used to convey agreement and to answer in the affirmative to a yes/no question, at 14 and 20 months. By 32 months, both head shakes and head nods were most often used in response to mothers’ yes/no questions (35.7% and 48.3%, respectively). Thus, over time, responding to yes/no question prompts replaced refusing and agreeing to do mothers’ suggested acts as the predominant reason toddlers were observed using head gestures.

Other communicative uses of head gestures included shaking the head ‘no’ to confirm a negative response to a yes/no question (i.e., double negative) and prohibiting the mother from performing an act. Thus, some of children’s head shakes were used as affirmative responses (e.g., Mother asks ‘It’s not a power saw?’, child simultaneously says ‘no’ and shakes head). In the latter case, a 20-month-old girl used a head shake while saying ‘no no no’, when her mother began to turn the crank of a jack-in-the-box. This act differs from a refusal in that it is directed towards the interlocutor’s actions rather than one’s own actions.

At each time point, a subset of head gestures was used for other reasons beyond these communicative acts. Some head shakes appeared to be used in contexts requiring self-prohibition (Pea, 1980; n = 3 instances; child stands near power outlet, shakes head ‘no’ and does not touch it), which may reflect the child’s use of the gesture for self-regulation, rather than for conveying a message to the parent. One head shake reflected the use of the word ‘no’ in a context involving a failed plan; child attempts to throw ball into box but misses repeatedly, mother says ‘almost … almost’ after each attempt, child shakes head and looks away, smiling (Gopnik, 1988; Gopnik & Meltzoff, 1985). In total, three head nods and one head shake appeared to be produced in error. In each case, the head nods were immediately followed by a head shake, as the child apparently intended to respond ‘no’. The mistaken head shake was made in response to the mother’s question, ‘what do you want me to color?’ The child paused, said ‘yeah’ and shook her head ‘no’.

Predicting verbal language development

To address the third research question, we used bivariate correlations and multiple regression modeling to examine relationships between children’s early head gesture production and the key outcome variable, pragmatic flexibility at Time 3. Control variables included features of early maternal language input, demographic factors, and children’s speech production at Time 1. Gesture production at 14 months was quantified in three ways for this analysis. First, a dichotomous variable indicated whether or not the child produced a head gesture for a communicative purpose at 14 months of age. Children received a score of 1 if they used at least one head gesture for one or more of six communicative functions: Refuse to do; Agree to do; Answer ‘no’ to a yes/no question, Answer ‘yes’ to a yes/no question; Make a statement with a negation; Agree with mother’s proposition. Children who did not use a head gesture for one of these purposes received a score of zero. The second version summed the number of these communicative intents the child conveyed via head gesture (max. = 6). The third version reflected head gesture frequency: the total number of head gestures produced by the child per minute of observation, times 10 (gestures per 10 minutes) at 14 months (Table 3). For this version, all instances of head gestures were counted, regardless of their function.

Control variables included child age at Time 3, child gender, Hollingshead score, Time 1 child word types and word tokens produced in 10 minutes, and Time 1 child pragmatic flexibility. Additional control variables included measures of maternal communicative input at Time 1: type/token ratio in verbal speech and the number of word tokens produced in 10 minutes.

Pearson bivariate correlations are presented in Table 4. Scatterplots of each predictor variable against the outcome measure were examined, and the appropriateness of examining linear relationships between the outcome and each variable was confirmed. Based on this analysis, the outcome measure of child pragmatic flexibility at Time 3 (right column) was positively associated with the child’s use of at least one head gesture as a communicative act at Time 1 (r = .442, p = .01) and with mother’s type/token ratio at Time 1 (r = .370, p = .034). The total number of communicative acts conveyed via head gesture had a weaker positive relationship with Time 3 pragmatic flexibility (r = .325, p = .065).

Table 4.

Bivariate correlations between child and mother communication measures and demographic variables.

	Demographics		T1 Child Communication						T1 Mother Language		T3 Pragmatic Flexibility
	2	3	4	5	6	7	8	9	10	11
Demographics
1. Age at Time 3	−.059	.141	.254	.297~	.290	.267	.277	.219	.256	.136	−.099
2. Gender (Female = 1; Male = 0)	–	.099	−.162	.140	−.065	.020	−.144	−.115	.149	−.105	.117
3. Hollingshead Score		–	.089	.050	.071	−.203	−.072	−.120	.212	−.346~	.219
Time 1 Child communication
4. Pragmatic Flexibility			–	.692***	.653***	−.035	.135	.049	.110	−.062	.044
5. Word types in 10 min.				–	.889***	−.049	.059	−.016	.160	.132	.092
6. Word tokens in 10 min.					–	.149	.273	.229	.139	.207	.104
7. Head Gesture Occurrence						–	.641***	.795***	.150	.380*	.442*
8. Head Gesture Frequency							–	.880***	.095	.114	.177
9. Head Gesture # Comm. Acts								–	.008	.154	.325~
Time 1 Mother language
10. Mother type/token ratio										.213	.370*
11. Mother word tokens										–	.189

Note: The numbers 2–11 across the top of the table correspond to the measures listed in the first column.

p < .10; *p < .05; **p < .01; ***p < .001.

Regarding child communication measures at Time 1, the three child gesture measures were highly correlated. Thus, these measures were not included simultaneously in regression models, due to possible collinearity problems. The same was true of the word type, word token, and early pragmatic flexibility measures. The only significant relationship between Time 1 mother and child communication was a positive association between mother word tokens and child head gesture occurrence (r = .380, p = .032). In this sample, gender and Hollingshead scores were not significantly correlated with any of the child or mother communication measures (all ps > .05). Nonetheless, these control variables were included in regression models, given their previously established relevance to children’s communication.

A series of multiple regression models was fit to the data, aimed at predicting variation in pragmatic flexibility at Time 3. Each model controls for the variation in child age at Time 3, and in session duration at Time 3, as pragmatic flexibility was calculated based on the complete communication sample. For the same reason, session duration at Time 1 was controlled in models that included pragmatic flexibility, gesture occurrence, and total gesture function scores at Time 1. Initial models confirmed the non-significant relationships between both gender and family Hollingshead scores and language outcomes (models not shown; parameter estimates p > .10). It is possible that the non-significant relationship between family socio-economic status and child language is due to a lack of socio-economic diversity in the current sample.

Model A (Table 5) presents the relationship between pragmatic flexibility at Time 1 and Time 3. This analysis confirms that pragmatic flexibility at 14 months is not a significant predictor of pragmatic flexibility at 32 months. Similarly, the number of word types children produced at Time 1 does not significantly predict pragmatic flexibility at Time 3 (model not shown, parameter estimate p > .10). The number of word tokens produced at Time 1 had a positive but weak association with Time 3 pragmatic flexibility (Model B; word token parameter estimate p = .06). For subsequent analyses, number of word tokens in 10 minutes was retained in the regression model as the measure of early language skill.

Table 5.

Results of fitting a taxonomy of multiple regression models predicting pragmatic flexibility at 32 months.

	Model A	Model B	Model C	Model D	Model E	Model F	Model G
Intercept	.51 (19.33)	13.57 (18.39)	21.49 (16.54)	17.17 (17.55)	13.87 (17.45)	21.87 (17.51)	19.10 (15.54)
Child age in months (32 months)	−.01 (.57)	−.40 (.56)	−.73 (.50)	−.61 (.55)	−.51 (.53)	−.77 (.52)	−1.01* (.49)
Session length (14 months)	.54~ (.28)		.22 (.24)		.32 (.25)	.22 (.25)	.10 (.23)
Session length (32 months)	.88* (.39)	1.27** (.37)	1.11** (.34)	1.42*** (.36)	1.11** (.36)	1.11** (.35)	1.07** (.32)
Child pragmatic flexibility (14 months)	−.10 (.36)
Child word tokens in 10 min (14 months)		.27~ (.13)	.21~ (.12)	.22~ (.13)	.17 (.13)	.19 (.13)	.20~ (.11)
Child head gesture occurrence (14 months)			5.39** (1.70)			5.52* (1.97)	5.32** (1.59)
Child head gesture frequency (14 months)				2.29~(1.13)
Child head gesture functions (14 months)					1.97* (.89)
Mother word tokens in 10 minutes (14 months)						.002 (.005)
Mother type/token ratio (14 months)							38.02* (17.45)
R²	0.33	0.33	0.55	0.42	0.47	0.56	0.62

p < .10; *p < .05; **p < .01; ***p < .001.

The addition of child communicative head gesture occurrence (Model C) yields a model that explains over half of the variance in Time 3 pragmatic flexibility (R² = 54.8%). Models D and E substitute head gesture occurrence with head gesture frequency (Model D) and number of communicative functions (Model E). Model D reveals that frequency is a weaker predictor, with a parameter estimate significant at the p < .10 level. Model E suggests that the number of communicative functions conveyed via head gestures significantly predicts pragmatic flexibility at Time 3. Comparing the R² values for Models C and E, head gesture occurrence explains more variability in pragmatic flexibility than does the gesture function variable. Thus, gesture occurrence was retained as the measure of early gesture use.

Building off of Model C, Models F and G add the main effects of maternal verbal input at Time 1. In each case, the parameter estimate for head gesture occurrence remains significant. Mother type/token ratio explains additional variance in the outcome (R² = 61.8%) and was thus retained in the final model (Model G). We also confirmed that the effect of head gesture occurrence (Model G) remains the same regardless of the measure of Time 1 child language used as a control.

Potential interaction effects were tested to examine whether the effect of head gesture occurrence varied by each of the other parent and child control variables. In each case, one interaction term (e.g., gender x head gesture occurrence) and its constituent main effects (i.e., gender) were added to Model G. None of these interaction terms were statistically significant at p < .05. However, the statistical power to detect interactions was limited by the small sample size.

Model G thus represents the final, best fitting model. Above and beyond the effects of early verbal language skill and maternal language input, the use of a communicative head gesture at 14 months predicts variation in the pragmatic flexibility of children’s verbal language at 32 months. Specifically, expressing at least one communicative intent through head gesture at 14 months is associated with expressing an average of five additional social interchange-speech act combinations at 32 months.

Discussion

In the context of parent-child play, head gestures occurred with increasing frequency in children’s communication from the second to third year of life. At 14 months of age, soon after head gestures typically first emerge, a minority of children were observed using them at all. By 32 months, most children were observed using head nods and head shakes, and with a greater frequency than observed at earlier time points. This result supports the notion that gaining control over the use of these gestures is a protracted developmental process, typically occurring in the second to third year of life.

Because conventional gestures can carry meaning on their own, children may produce head gestures in isolation, with non-word vocalizations, or in combination with verbal utterances. Developmental change was observed in the proportion of head gestures produced with and without a verbal message. When children produced head nods, they most often did so in the absence of a verbal utterance, though the proportion of speech+gesture combinations increased over time. The same was true of head shakes, to the point where, at 32 months, the majority of children’s headshakes were accompanied by verbal utterances. Thus, during this developmental period, children begin to coordinate verbal messages with head gestures. It is noteworthy that this representational control seems to be achieved earlier with head shakes, which also emerge earlier than head nods in children’s communicative repertoire (Fenson et al., 1994). Along these same lines, we observed a slight increase in average frequency rates for head shakes between 14 and 20 months, and a sharp increase in head nodding rates in the later and longer gap between 20 and 32 months. A future study might probe possible cognitive and contextual explanations for the relatively protracted mastery of head nods, and their integration with speech, in children’s spontaneous communication.

Head nodding rates were somewhat higher than head shaking rates at 32 months. Although this pattern replicates Guidetti (2005), in the current study the trend was not statistically significant. Lack of statistical significance in this case may be due to the wide individual differences in head gesture frequency rates, and the small sample size. Interestingly, based on a separate analysis (Fusaro, 2009), the mothers from these dyads consistently produced more head nods than head shakes during the play sessions. Further research would be needed to understand the potential contextual and social/emotional reasons that head nods might eventually overshadow head shakes in mother-toddler play. It may be that, by 32 months, parent-child dyads begin to settle into interaction patterns during play that are biased towards positivity. However, this pattern may not emerge among dyads that differ from the current sample, including those in different cultural settings or those interacting under more stressful conditions. Indeed, the relative frequency of head nods versus head shakes in dyadic interactions may provide insights into social and emotional aspects of the parent-child relationship. For example, dyads that use few head nods or many head shakes in a setting designed to be playful may experience less harmonious interactions in other contexts as well (Grimm, Vallotton, & Ayoub, 2009).

In line with previous work emphasizing both social pragmatic and cognitive underpinnings of language, refusal was the predominant communicative intent conveyed via the earliest head shaking gestures (Carpenter et al., 1983; Crais et al., 2004; Gopnik & Meltzoff, 1985; Pea, 1980). Refusal is a relatively simple message, in that it can be produced somewhat independently of the verbal discourse context (Carpenter et al., 1983). As Pea (1980) has argued, a refusal can be made in response to an immediately perceptible experience, rather than as a comment about a preceding linguistic utterance. Interestingly, the counterpart to refusing, namely, agreeing to do an act proposed by the mother, was also observed at 14 months. Agreeing and refusing to do an activity may be relatively early emerging intentions because they each operate as direct responses to another person’s actions, rather than to the verbal messages that might accompany those actions.

By the third time point, refusals were less predominant in children’s total head gesture repertoire. Several factors likely contribute to this shift. Of course, as children gain control over additional communicative intents, refusals will account for a diminishing proportion of communicative acts. Further, children likely conveyed some refusals with only verbal language by 32 months (Guidetti, 2005). Indeed, an earlier study using the New England CHILDES sample reported that children’s general reliance on non-verbal forms to convey their communicative intents decreased over this developmental period (Pan et al., 1996). However, a reduction in refusal messages may also be explained by a change in mother-child interactions over time. Namely, over time, mothers may better anticipate which activities or situations their child is likely to protest, and make adjustments to avoid conflict (e.g., Carpenter et al., 1983). Further, when children are faced with something undesirable they may be better able to regulate their reaction at 20 and 32 months than at 14 months, which may correspond to a decrease in non-verbal refusal messages.

The current study extends examinations like those by Carpenter and colleagues (1983) that found that answering begins to emerge by 15 months. As those authors argued, to use conventional forms of communication to answer, such as in response to yes/no questions, children must coordinate their communication with that of the other speaker. Pan and colleagues (1996) raise a similar point in describing the relatively late emergence of agreement and disagreement messages in children’s verbal responses to mothers’ propositions, namely, that control of these speech acts reflects emerging abilities to operate at the discourse level. In this study, answering yes/no questions became the predominant use of head nods and head shakes by the third time point. Children’s use of head gestures in response to mothers’ questions and comments, particularly in the absence of speech, highlights the significance of including gestures when examining pragmatic development beyond the first year of life.

Nodding the head in agreement with propositions, another form of answering, was also observed in a greater number of cases at Time 3. Interestingly, children were not observed using head shakes to disagree with their mothers’ claims. Because we examined spontaneous communication, we cannot determine whether conveying disagreement with a head gesture emerges relatively late compared to agreement messages, or whether this communicative function is less relevant to the child’s goals in this play context. In both verbal and non-verbal modalities, disagreement would be elicited if the mother spontaneously made a statement that the child might question, or if the child was motivated to disagree. These conditions may be relatively rare in this particular context. Additional studies aimed at eliciting agreement and disagreement messages would shed more light on the emergence of this non-verbal component of discourse competence.

Somewhat surprisingly, as a group, the children used head gestures for a similar range of communicative functions at 14, 20 and 32 months of age. Rather than in the breadth of observed functions, increases were observed in the number of children observed producing head gestures in the service of this core set of functions. Based on a related study of head gesture production by mothers in these same dyads, mothers used head gestures for a broader range of communicative functions than children did (Fusaro, 2009). For example, mothers nodded in approval of certain behaviors, and shook the head to disapprove of other behaviors. Because communicative messages are used in the service of achieving certain goals, it is possible that the absence of some gesture functions in children’s communication reflects their irrelevance to the child’s goals; mothers, unlike children, intend to socialize children’s behavior by promoting valued and prohibiting non-valued behaviors. Further, interactions with peers might elicit alternative uses of head shakes, such as disagreeing with another child’s statements. Again, a future study that elicits a range of communicative responses would help to identify when children are able to express particular communicative intents, and whether these skills emerge earlier in gesture than speech.

Infants’ communicative use of head gestures predicts higher levels of pragmatic flexibility in verbal language as toddlers, even after accounting for early verbal language skills and maternal speech input. That is, children who used head gestures at 14 months to agree or refuse to do acts, and to respond to mothers’ questions and propositions tended to have greater command over the social uses of verbal language by 32 months. To explain this relationship, we must consider the continuities between head gesture use and language. By definition, intentional communication through gestures and words involves the deliberate use of conventional signals to convey intent and potentially affect another person’s behavior (Bates et al., 1975). Perhaps the early gesture users have achieved an earlier command of conventional forms of communication, compared to the non-users; a head shake conveys the same message as pushing a peek-a-boo cloth away, but does so in a way that meets the mother at the level of representation (or coordinated systems of actions) rather than with less complex actions (Fischer & Bidell, 2006; Fischer & Corrigan, 1981). Those children who have a grasp of these non-verbal conventional signals may have the command of conventional verbal signals within their zone of proximal development (Vygotsky, 1978). Further, it is possible that those children who are early to use head gestures have a better understanding of the influence of their communicative signals on the behavior of communication partners. Thus, differences in social understanding may be driving the predictive effect of early gestural communication and later pragmatic skill.

Another related explanation of the association between early head gesture and later language skill draws on the notion that early gesture use, particularly a response to mothers’ words, reflects an emerging grasp of the verbal discourse context. Children must construct an understanding of the turn-taking nature of discourse. This skill may have roots in the earliest caregiver-child face-to-face interactions (Snow, 1977). Early proto-conversations may reflect coordinated communication at the level of vocalizations and emotional expression. Again, reconstructing this skill at the level of turn-taking with conventionalized signals may be a watershed for continued advances in verbal language skills. Those children who demonstrate advances in discourse skills through their use of head gestures may be in a good position to use words more flexibly in dyadic interaction with their caregivers. Further, early gestures may not only precede verbal language development, but may also facilitate it (Sauer et al., 2010). The option of using a non-verbal head gesture instead of speech may provide the child with a less demanding means for engaging effectively in conversational turn-taking. Thus, the child may build skills in basic conversational routines, like yes/no questioning and acknowledging another person’s messages, by relying initially on gestural acts.

This examination of young children’s production of two conventional gestures provides useful insights into several aspects of emerging communication skills. Change over time in the frequency of head gesture production indicates that learning to use head gestures in response to appropriate antecedents is a developmental challenge for typically developing children in the second year of life. As with other co-speech gestures, children also need to learn how to coordinate head nods and head shakes with verbal utterances in order to produce fluid multimodal messages. The period between 14 and approximately 32 months encompassed the emergence of isolated head nods and head shakes, as well as head shakes paired with verbal speech, for most children. At some point beyond this range, children master the coordination of head nods and verbal speech as well; additional studies might examine just how protracted this process is in typical development. Further, the early use of head gestures in parent-child communication is predictive of additional progress in the pragmatics of verbal language. This relationship provides additional support for the view that pre-verbal and verbal communication rely on overlapping social and cognitive systems. Taking this assumption a step further, head gestures used in turn-taking exchanges may play a facilitative role in the emergence of discourse skills in verbal language. Finally, the interaction context was highly relevant, as the dyadic play setting provided both opportunities and constraints for the child’s involvement in communicative exchanges. Continued analysis of children’s spontaneous gesture use in varied contexts, and prompted use in response to elicitation, will further illuminate the process by which children enter into reciprocal, multimodal conversation with others.

Footnotes

Acknowledgements

This work is dedicated to our late collaborator, Barbara Alexander Pan. We would also like to acknowledge several graduate students who assisted with data coding: Lindsay Fryer, Julia Hayden, Dayna Johnson, Sarah Lipson, and Rose Wongsarnpigoon.

Authors’ Note

Maria Fusaro, Human Development and Psychology, Harvard Graduate School of Education. Paul L. Harris, Human Development and Psychology, Harvard Graduate School of Education. Barbara A. Pan, Human Development and Psychology, Harvard Graduate School of Education. Maria Fusaro is now at the M.I.N.D. Institute, University of California, at Davis. Data from this report were included in a doctoral dissertation submitted by Maria Fusaro to the Graduate School of Education of Harvard University.

Funding

The first author was supported by the Charles H. Smith Scholarship fund.

References

Acredolo

L. P.

Goodwyn

S. W.

(1988). Symbolic gesturing in normal infants. Child Development, 59, 450–466.

Bates

Benigni

Bretherton

Camaioni

Volterra

(1979). Cognition and communication from nine to thirteen months: Correlational findings. In Bates

(Ed.), The emergence of symbols: Cognition and communication in infancy (pp. 69–140). New York: Academic Press.

Bates

Camaioni

Volterra

(1975). The acquisition of performatives prior to speech. Merrill-Palmer Quarterly, 21, 205–226.

Bruner

(1981). The social context of language acquisition. Language and Communication, 1, 155–178.

Capirci

Iverson

J. M.

Pizzuto

Volterra

(1996). Gestures and words during the transition to two-word speech. Journal of Child Language, 3, 645–675.

Carpenter

Nagell

Tomasello

Butterworth

Moore

(1998). Social cognition, joint attention, and communicative competence from 9 to 15 months of age. Monographs of the Society for Research in Child Development, 63(4, Serial No. 255).

Carpenter

R. L.

Mastergeorge

A. M.

Coggins

T. E.

(1983). The acquisition of communicative intentions in infants eight to fifteen months of age. Language and Speech, 26, 101–116.

Crais

Douglas

D. D.

Campbell

C. C.

(2004). The intersection of the development of gestures and intentionality. Journal of Speech, Language, and Hearing Research, 47, 678–694.

Fenson

Dale

Reznick

J. S.

Bates

Thal

Pethick

(1994). Variability in early communicative development. Monographs for the Society for Research in Child Development, 59(5, Serial No. 242).

10.

Fischer

K. W.

Bidell

T. R.

(2006). Dynamic development of action and thought. In Damon

Lerner

R. M.

(Eds), Theoretical models of human development. Handbook of child psychology (6th ed., Vol. 1, pp. 313–399). Hoboken, NJ: Wiley.

11.

Fischer

K. W.

Corrigan

(1981). A skill approach to language development. In Stark

R. E.

(Ed.), Language behavior in infancy and early childhood (pp. 245–273). North Holland: Elsevier.

12.

Fusaro

(2009). Toddlers get the nod: An analysis of ‘yes’ and ‘no’ gesture use during the second to third year of life. Ed.D. dissertation. Harvard University, MA. Dissertations & Theses: A&I. (Publication No. AAT 3393909).

13.

Gopnik

(1988). Three types of early word: The emergence of social words, names and cognitive-relational words in the one-word stage and their relation to cognitive development. First Language, 8, 49–70.

14.

Gopnik

Meltzoff

A. N.

(1985). From people, to plans, to objects: Changes in the meaning of early words and their relation to cognitive development. Journal of Pragmatics, 9, 495–512.

15.

Grimm

Vallotton

C. D.

Ayoub

(2009). Experience mirrored in language and gesture: Maltreated children’s use of grammatical negations and head shakes. Poster presented at MultiMod 2009: Multimodality of communication in children, Toulouse, France.

16.

Guidetti

(2005). Yes or no? How young French children combine gestures and speech to agree and refuse. Journal of Child Language, 32, 911–924.

17.

Hollingshead

(1975). The four-factor index of social status. Unpublished manuscript. Department of Sociology, Yale University, New Haven, CT.

18.

Iverson

J. M.

Capirci

Longobardi

Caselli

M. C.

(1999). Gesturing in mother-child interactions. Cognitive Development, 14, 57–75.

19.

Iverson

J. M.

Capirci

Volterra

Goldin-Meadow

(2008). Learning to talk in a gesture-rich world: Early communication in Italian vs. American children. First Language, 28, 164–181.

20.

Iverson

J. M.

Goldin-Meadow

(2005). Gesture paves the way for language development. Psychological Science, 16, 368–371.

21.

Kratochwill

T. R.

Wetzel

R. J.

(1977). Observer agreement, credibility and judgment: Some considerations in presenting observer agreement data. Journal of Applied Behavior Analysis, 10, 133–139.

22.

Laakso

M. L.

Poikkeus

A. M.

Katajamäki

Lyytinen

(1999). Early intentional communication as a predictor of language development in young toddlers. First Language, 19, 207–231.

23.

Liszkowski

Carpenter

Striano

Tomasello

(2006). Twelve- and 18-month-olds point to provide information for others. Journal of Cognition and Development, 7, 173–187.

24.

MacWhinney

(2000). The CHILDES Project: Tools for analyzing talk. Mahwah, NJ: Erlbaum.

25.

MacWhinney

Snow

C. E.

(1985). The child language data exchange system. Journal of Child Language, 12, 271–295.

26.

MacWhinney

Snow

C. E.

(1990). The child language data exchange system: An update. Journal of Child Language, 17, 457–472.

27.

Morford

Goldin-Meadow

(1992). Comprehension and production of gesture in combination with speech in one-word speakers. Journal of Child Language, 19, 559–580.

28.

Ninio

Snow

C. E.

(1996). Pragmatic development. Boulder, CO: Westview Press.

29.

Ninio

Snow

C. E.

Pan

B. A.

Rollins

P. R.

(1994). Classifying communicative acts in children’s interactions. Journal of Communication Disorders, 27, 157–187.

30.

Ninio

Wheeler

Snow

C. E.

Pan

B. A.

Rollins

P. R.

(1991). INCA-A: Inventory of Communicative Acts – Abridged. Unpublished manuscript. Harvard Graduate School of Education, Cambridge, MA.

31.

Pan

B. A.

Imbens-Bailey

Winner

Snow

C. E.

(1996). Communicative intents of parents interacting with their young children. Merrill-Palmer Quarterly, 42, 248–266.

32.

Pea

R. D.

(1980). The development of negation in early child language. In Olson

D. R.

(Ed.), The social foundations of language and thought: Essays in honor of Jerome S. Bruner (pp. 156–186). New York: W.W. Norton.

33.

Rowe

M. L.

Goldin-Meadow

(2009). Early gesture selectively predicts later language learning. Developmental Science, 12, 182–187.

34.

Rowe

M. L.

Özçalışkan

Goldin-Meadow

(2008). Learning words by hand: Gesture’s role in predicting vocabulary development. First Language, 28, 182–199.

35.

Sauer

Levine

S. C.

Goldin-Meadow

(2010). Early gesture predicts language delay in children with pre- or perinatal brain lesions. Child Development, 81, 528–539.

36.

Sears

R. R.

Rau

L. L.

Alpert

(1965). Identification and child rearing. Stanford, CA: Stanford University Press.

37.

Snow

C. E.

(1977). The development of conversation between mothers and babies. Journal of Child Language, 4, 1–22.

38.

Snow

C. E.

Pan

B. A.

Imbens-Bailey

Herman

(1996). Learning how to say what one means: A longitudinal study of children’s speech act use. Social Development, 5, 56–84.

39.

Tomasello

Carpenter

Liszkowski

(2007). A new look at infant pointing. Child Development, 78, 705–722.

40.

Vygotsky

L. S.

(1978). Mind in society: The development of higher psychological processes. Cambridge, MA: Harvard University Press.

41.

Wetherby

A. M.

Cain

D. H.

Yonclas

D. C.

Walker

V. G.

(1988). Analysis of intentional communication of normal children from the prelinguistic to the multiword stage. Journal of Speech and Hearing Research, 31, 240–252.

42.

Yont

K. M.

Snow

C. E.

Vernon-Feagans

(2001). Early communicative intents expressed by 12-month-old children with and without chronic otitis media. First Language, 21, 265–287.